Editable PDF to Word

A minimal Python tool for converting text-based PDF files into editable DOCX documents. It uses pdf2docx to reconstruct the layout, PyMuPDF to inspect and validate the PDF, FontTools to inspect embedded fonts, and python-docx to fix common image and heading issues after conversion.

Features

• Converts PDF text, images, tables, and page breaks into an editable DOCX.
• Preserves text formatting and embeds permitted fonts found in the PDF.
• Fixes common image-spacing and bullet-font issues.
• Optionally assigns Word Heading styles and repairs short fallback-font runs.
• Runs locally without uploading files or downloading fonts.

Requirements

• Python 3.10 or later; Python 3.12 is recommended.
• A text-based PDF. Scanned PDFs require an OCR step before conversion.

Installation

Clone the repository and enter the project directory:

git clone https://github.com/nguyennt06/pdf_to_word.git
cd pdf_to_word

Install the dependencies once.

Linux, macOS, or WSL:

python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install --upgrade pip && python3 -m pip install -r requirements.txt

Windows PowerShell:

py -3 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip; if ($LASTEXITCODE -eq 0) { python -m pip install -r requirements.txt }

Usage

Keep the virtual environment active and run:

python3 main.py input.pdf

Choose the output file:

python3 main.py input.pdf -o output.docx

Input and output files may be anywhere on the filesystem. Quote paths that contain spaces:

python3 main.py "/path/to/input file.pdf" -o "/path/to/output file.docx"

Detect numbered headings and assign native Word Heading styles:

python3 main.py input.pdf -o output.docx --headings

Overwrite an existing output file:

python3 main.py input.pdf -o output.docx --force

Keep font names but create a smaller DOCX without embedded PDF fonts:

python3 main.py input.pdf -o output.docx --no-embed-fonts

On native Windows, replace python3 in the usage examples with python after activating the virtual environment.

Font handling

• Font families are recovered from PDF metadata without hard-coded Calibri or Arial replacements. Size, weight, italic, color, script-specific fonts, and Heading formatting are preserved when available.
• Embedded TrueType/OpenType fonts are copied from the PDF only when their license flags permit it. Missing, restricted, or invalid fonts produce a Font warning; the tool never searches for or downloads fonts.
• PDF fonts are often subsets. If a normalized heading needs missing glyphs, the incomplete subset is skipped and Word uses the installed full font.
• Use --no-embed-fonts to keep font names without including font files in the DOCX. Word may substitute fonts that are not installed.

Troubleshooting

If Python reports No module named ..., install the dependencies with the same Python interpreter used to run the tool.

With the project's virtual environment:

source .venv/bin/activate
python3 -m pip install -r requirements.txt

Without a virtual environment:

python3 -m pip install --user -r requirements.txt

Use a .docx extension when setting the output path.

Limitations

• DOCX reflows content, so page layout and detected headings should be reviewed.
• Scanned PDFs require OCR; complex equations, columns, and tables may need manual correction.
• Fonts cannot be embedded when they are missing from the PDF or their license flags prohibit it.

License

Released under the MIT License.